A computer that learns to talk in the same way as a young child by holding conversations with humans has been developed by scientists. The machine, which uses cutting edge artificial neural network technology to mimic the way the human brain works, was given 1,500 sentences from literature about language structure. It was then able to use this to learn how to construct new sentences with nouns, verbs, adjectives and pronouns when having a conversation with a real human. Researchers used connections between two million artificial neurons to mimic some of the processes that take place in the human brain as we learn to speak (illustrated). While some of the sentences had the rather functional approach of a computer rather than the finesse of a natural speaker, the results are still impressive.

It is a technique used to help athletes achieve their goals and is thought to be how young children learn to walk, but it appears robots can also use visualisation to help them move around. Robots engineers have developed simulated neural networks that allow a toddler-sized robot imagine how it can stand up and move around before it attempts the task. They have been using the technique to help teach the little humanoid robot, nicknamed Darwin, to learn how to stand on its own two feet. Darwin, a toddler sized robot (pictured), is being taught how to stand, lean and move in much the same way as a young child by first getting'imagining' how it might perform the movement through a series of simulations The process is designed to mimic the way young children first start learning to walk - with a great deal of trial and error. It is intended as a way of overcoming some of the problems engineers have faced in producing humanoid robots that do not fall on their face.

Scientists have revealed an artificial intelligence (AI) system that can solve SAT geometry questions as well as the average American 11th-grade student. Called GeoS, it uses a combination of computer vision to interpret diagrams, natural language processing to read and understand text and a geometric solver to achieve 49 percent accuracy on official SAT test questions. If these results were extrapolated to the entire Math SAT test, the computer achieved an SAT score of 500 (out of 800), the average test score for 2015, the team behind it say. The system uses a combination of computer vision, natural language processing and a geometric solver to achieve 49 percent accuracy on official SAT test questions. GeoS is the first end-to-end system that solves SAT plane geometry problems.

Protracted negotiations within the United Nations are causing'killer robots' to move a step closer to reality, an expert has warned. The UK and US are currently attempting to water down a preemptive ban on autonomous weapons at the UN general assembly in New York. But this is delaying an agreement on banning the technology, which could allow nations to possess killer robots before any laws come into force. Protracted negotiations within the United Nations are causing'killer robots' to move a step closer to reality, an expert has warned. This was the stark warning made by Christof Heyns, the UN special rapporteur on extrajudicial, summary or arbitrary executions, in an in-depth article for The Guardian.

It's hard to think of mass extinctions as positive events, with species and genetic material lost forever. But computer scientists have shown that mass extinctions such as the one that killed off the dinosaurs could actually speed up evolution, leading to useful new adaptations – in robots at least. They say the finding has implications for the development of artificially intelligent robots and may support some evolutionary biologists' views that extinction events actually promote lineages that have the biggest potential to evolve. 'Focused destruction can lead to surprising outcomes,' said Risto Miikkulainen, a professor of computer science at the University of Texas at Austin. 'Sometimes you have to develop something that seems objectively worse in order to develop the tools you need to get better.'

The ability to make decisions is seen as one of the most important intellectual abilities for humans and animals. Now Japanese scientists claim that an inanimate objects such as an iron bar is capable of gathering information about its surroundings to make a'decision' too. They believe their study has implications for understanding how lower life forms function. Japanese scientists claim that inanimate objects such as an iron bar are capable of gathering information about their surroundings to make a decision. Song-Ju Kim, of National Institute for Materials Science, Tsukuba, told Phys.org: 'The most important implication that we wish to claim is that the proposed scheme will provide a new perspective for understanding the information-processing principles of certain lower forms of life.

Researchers have found a Turkish language based on whistles is unique as it uses both sides of the brain. Generally speaking, language processing is a job for the brain's left hemisphere, whether it is spoken, written, or signed. However, researchers have discovered an exception to this rule: whistled Turkish. Whistled Turkish is Turkish that has been adapted into a series of whistles. This method of communicating was popular before the advent of telephones, and can travel up to 90m away.

Artificial intelligence could be a reality within years, rather than decades, scientists have said. Elon Musk recently warned AI has the potential to be as dangerous to mankind as nuclear weapons. But is there hope for artificial intelligence in the form of religion? And could faith allow AI to do more good than harm? Elon Musk recently warned AI has the potential to be as dangerous to mankind as nuclear weapons.

Before you pull on your loafers, double denim and flat cap, you may want to turn to technology for advice. Scientists have created a mathematical model that can give people fashion tips by analysing the latest season trends. The system uses an algorithm to provide a score on how fashionable your outfit or tell you if you're a walking disaster. Scientists have created a computer that can give people fashion tips by analysing the latest season trends. The technology was developed by Spain's Institute of Robotics and Industrial Informatics, along with colleagues at the University of Toronto.

Google's DeepMind division is using Daily Mail and CNN articles to teach its artificial intelligence programs to read. Using the unique style of articles on the sites - with concise bullet points summarising a story at the top of a page - artificial intelligence was able to learn key facts about articles to answer queries. Ultimately, scientists hope that the study could lead to complex artificial'brains' that can read entire documents and respond to questions put to them by a human. The British-based DeepMind unit analysed almost 400,000 articles from the sites (language process shown). They were used for their unique style of bullets, text and captions. Artificial intelligence was able to learn key facts from the articles.